Unable to execute direct_method calls when first connect() gets called before device is known in the IoT-Hub

Context

• OS and version used: <VERSION> (Ubuntu 18)
• Python version: <VERSION> (3.7.5)
• pip version: <VERSION> (pip 9.0.1)
• cloned repo: <VERSION> (azure-iot-device==2.1.0)

Description of the issue:

Unable to execute direct_method calls when first connect() gets called before device is known in the IoT-Hub. Steps to reproduce:

• make sure the device it not added yet to the IoT-Hub
• create the IoTDeviceClient object
• call connect() This will fail with a: CredentialError(‘Credentials invalid, could not connect’) caused by UnauthorizedError(‘Connection Refused: not authorised.’)
• Add the device to the IoT-Hub
• call connect() for the second time ON THE SAME INSTANCE of the IoTDeviceClient.
• connection succeeds --> async_clients(110): connect: Successfully connected to Hub
• Try to execute a direct method call -> This doens’t work (not in the web portal and not via the azure cli tool)
• Waited for the sas_token_update, but even after that, still not possible to send direct_method calls to the device.

Response when executing a direct mothod call directly from the IoTHub web portal: {“message”:“Device {"Message":"{\"errorCode\":404103,\"trackingId\":\"7af44c59b01e49539bb0dc91cc717798-G:21-TimeStamp:03/29/2020 10:47:23-G:6-TimeStamp:03/29/2020 10:47:23\",\"message\":\"Timed out waiting for device to subscribe.\",\"info\":{},\"timestampUtc\":\"2020-03-29T10:47:23.0351741Z\"}","ExceptionMessage":""} not registered”}

I have a task monitoring the IoTHubDeviceClient to recover from cases where the device is not known in the IoTHub for whatever reason. If the device gets again added to the IoT-Hub, after been removed by accident, we should be able to recover without rebooting. This isn’t the case for now I think. Connect indeed succeeds when the device is added again, but direct method calls are not working anymore. I know the SDK tries to reconnect on its own when the connection drops. But this is a monitoring task to try to cover all other cases where for instance a mistake is made in the cloud and the IoT-Devices get removed by accident. In that case, you want the device to recover on its own.

Code sample exhibiting the issue:

     while True:
            if not iot_hub_device_client.connected:
                LOG.debug("iot_hub_device_client is not connected, call connect()")
                # Connect IoTHubDeviceClient
                try:
                    await iot_hub_device_client.connect()
                except Exception as exception:
                    message = "%s: %s" % (type(exception).__name__, exception)
                    LOG.error(message)
            else:
                LOG.debug("iot_hub_device_client is still connected")
            await asyncio.sleep(60)

Console log of the issue:

logfile.txt

AB#7366702

[BertKleewein]

fixed in azure-iot-device 2.1.2 – https://pypi.org/project/azure-iot-device/2.1.2/

[BertKleewein]

I don’t know what the version number will be, but it will be in the next release. I’m trying to figure out when that will be. Hopefully soon.